Security of Neural Networks
Defending deep models against adversarial examples, backdoor and data-poisoning attacks — and understanding when and why they fail.
Research studies, software and prototypes from across the NPerseus team.
Research studies, prototypes and software described in public profiles. Status, roles and funding follow each source record; a profile entry does not by itself establish a released product.
74 matching profile project records · Page 2 of 7
developed a voice-driven desktop assistant integrating automation and information-retrieval tasks.
a prototype integrating Excel, PDF and CSV operational data, train-readiness dashboards and train-induction scheduling.
the published study classifies Android malware using high-level behavioral features.
Built an attendance and organization-management platform with mobile and kiosk clients, an administrative dashboard and backend services.
investigates vulnerabilities in distilled datasets and develops poisoned-sample detection methods using feature-space analysis, layer-wise representation consistency, Mahalanobis distance and feature similarity.
public TensorFlow/Keras U-Net notebook for ultrasound-image segmentation.
M.Tech thesis work on prediction for privacy-preserving image data hiding.
Built ContextGuard, a device-access control project spanning mobile apps, an administrative interface and smart-contract components, following an earlier native Android app-blocking prototype.
TypeScript, React and Express prototype for exploring Ethereum wallet activity through transaction analysis, scoring rules and dashboards; not a validated credit-rating service.
Built experimental dataset-distillation software for data synthesis, evaluation and experiment tracking.
full-time Ph.D. research under Rajeev Kumar, with Anurag Goel as co-supervisor at DTU.
React and Node.js documentation prototype that analyses repository source, generates AI-assisted descriptions and publishes them to a GitHub wiki with progress updates.
Defending deep models against adversarial examples, backdoor and data-poisoning attacks — and understanding when and why they fail.
Reversible data hiding in encrypted images, image and text steganography, and robust watermarking for ownership and integrity.
Detecting manipulated and GAN-generated media — deepfakes, splicing and median-filtering traces.
Federated and privacy-preserving learning, explainable and trustworthy models, and methods for limiting the disclosure of training data.